Robert D. Davis
Great commentary by Chris Lamb (July 2005)
Hi! I just came across the video of the Intran/PERQ demo, and was blown away. I'm an avid collector of PERQs, and the original self-described "PERQ Fanatic," starting a mailing list (around 1990?) of the same name and eventually helping to create the alt.sys.perq newsgroup, contributing to the R.D. Davis PERQ FAQ, and sharing both hardware, stories and even software with Robert and the other major U.S. PERQ collector, Bill von Hagen. As a kid, I spent many, many hours hanging out at CMU in their "SPICE Rack" and at the Three Rivers Computer "Advanced Development Office" with my stepdad, who was both a CMU grad and employee of 3RCC in the early days. It was my stepdad who introduced me to programming on the PERQ, and it was and is my "first love" in computing. In short, I am a total nutball for the PERQ.
I currently own three systems, and am _constantly_ on the lookout - even at this late date, 20 years after the company died - for additional systems, parts, software, and stories from PERQ people. I'm stunned and saddened that so much of my life has been busy with "work" and other things to have much time to spend on my own machines; they are currently here in my basement, carefully covered and "mothballed" while I finish some remodelling work - but I hope _very soon_ to be able to get them back up and running. I plan to put together my own definitive PERQ web site, and if I can complete the work I started (ten YEARS ago!?) debugging the POS/Accent TCP/IP stack, I hope to even have a PERQ live on the Internet - possibly for the first time...?
I'll certainly be in touch as my own preservation efforts proceed - I'd love to swap links and images and other materials with you guys. I'm thrilled to see that you have a PERQ, and the Intran stuff is fascinating.
I also have one request: PLEASE, PLEASE consider contacting Al Kossow of "bitsavers.org" and letting him create archive copies of the Intran software, along with any other PERQ software and manuals you guys have. I shipped off two huge boxes of random PERQ floppies to Al, and he's been able to sort through them and create images in "dmk format" (explained on the bitsavers site) as well as PDFs of the documentation. You guys may already know about Al's collection of Alto hardware and software, and may already be aware of his excellent work to preserve older material through the bitsavers archive. I really hope that the Intran stuff could be added to the collection on-line, since it really was an amazing system (and I'd *leap* at a chance to play with the software on my own T2 someday!
So this note is mostly to say "Hi!" and let you know that there are still some PERQ nuts out here who really appreciate the work you guys are doing. I'll definitely be in touch as soon as the current craziness at home here settles down a bit and I can spend some quality time putting together my own PERQ pages. (I also own a pair of Cray CS6400's, which I hope to document along with my PERQ stuff under the domain "heavycomputing.com"
P.S. The picture you have captioned "Older ICL Perq" is actually the _original_ PERQ 1. The squatter, rounder cabinet used in the PERQ 2 (T1 & T2) was the ICL design; the funny shaped monitor and skinnier box with the brown plastic front and back panels was the first Pittsburgh-built PERQ. Apparently some used to jokingly call the machine the "super-duper plastic RasterOp machine" because of the funny monitor design and those plastic panels...
And more from Chris...
Curator: Did you know that the Mach kernel, core of todays Mac OSX, was developed on the PERQ?
Oh, definitely! I'm keenly aware of the MacOSX->NeXT/OpenStep->Accent
lineage. I think Accent is insanely cool, given how much they squeezed
into 16K of WCS... I've got a PERQ-1A with "F"-rev boards running POS F.2
and Accent S5 (with Lisp!), a PERQ-1A with "G"-rev boards running POS G.6
and Accent S6, and a PERQ-T2 with POS G.6/Accent S6 that I was using to
debug the TCP/IP software I got ahold of from Don Scelsza. I would kill
to get my hands on another pair of T2s - or a color board, a 24-bit board
set, an "MLO" board, the Audre camera/scanner gizmo...
One of the 1A's has some badness on its backplane (temporarily repaired by
wirewrapping the pins!) and its monitor is quite obviously burned in - it
was a CMU machine that I picked up from a friend who moved to Boston and
didn't want to lug it across PA. (In fact, I've found a replacement
source for a compatible CRT at Clinton Electronics in Illinois - if I
could just find a competent monitor repair guy here in town...)
My other 1A is in good shape; I just haven't run it in a while, and don't
recall any problems with it...
The PERQ T2
My T2 was actually shipped in beautiful Anvil cases, the last demo machine
saved when the holding company that held PERQ Systems' assets finally
unloaded them. Bill got a bunch of spare parts and stuff too, I think...
But, sadly, my machine was a little confused, last I left it... in my
excitement at finding a cheap Maxtor XT1140 to expand my disk space, I
misapplied the ID jumper and booted the machine with both drives sharing
the same ID. Doh! I was hoping it wouldn't be any big deal; shut down,
changed the jumpers, and tried to reboot... but corrupted something badly
on the boot drive and left the machine unable to boot from disk. So IIRC,
the T2 has to be rescued from floppy, and that process was slow going.
Trying to manually patch together the different Accent and POS boot blocks
using just floppy-based tools takes... patience.
But I *really* really really don't want to lose the data on that machine,
because it's the one that had the TCP/IP software on it. Argh! But it's
a good project to work on.
Curator: We have a couple of Crays here (1, and Q2) and a great Alto II/XM.
Crays are insanely neat, even the "pseudo-Cray" (SPARC-based) CS6400's
I've got. Damn they're cool.
And I loved the Alto, as a kid, because the games on it were phenomenal.
I was a MazeWar *fiend*. Pinball, and the Trek game, and the drawing
programs... but for me, the PERQs were sexier, because I knew a LOT more
about them, getting to hang out with Brian Rosen, John Strait, Don
Scelscza, Terry Vavra, Brad Meyers, and all those guys - I got to see the
prototype machines, see the first Ethernet-based time and print servers
(the $18,000 Canon LBP printer, wow!) and all the silly games and demos
and diversions and neat software that never made it out the door...
One of the most frustrating and difficult stories to relate about the PERQ
was that of the "VLSI PERQ" - essentially a single-board reimplementation
of the entire microengine, memory/display/DMA controller, and I/O board on
5 ASICs. They had all the designs taped out, "in the can" - and needed to
raise $1M, apparently, to get silicon. And the investors killed them, no
more VC money would be forthcoming. IDIOTS! The styrofoam mockups and
design drawings of the cabinetry were of a slender PERQ in what would have
looked like a DEC BA123(?) thin tower case, with a sleeker, less bulky CRT
design. Had the machine been produced, it could have made the PERQ a
_much_ more viable platform - under $10k retail pricetag (would have been
a first for the "workstation class" at the time?), 5.25" disk drives
(better performance), much lower manufacturing costs, greater reliability,
far less heat/power/noise, etc.
Just like the "NRW" (the legendary NeXT RISC Workstation, or "brick")
could have been the first commercial PowerPC machine -- how ironic THAT
would have been, Steve Jobs beating Apple & IBM to first shipment! -- the
VLSI PERQ is the coolest machine that never was. It could have possibly
saved the company.
Sadly, I learned from Brian Rosen that the PERQ-3, of which a few actually
did slip out of ICL (BillvH has a couple!), was going to be a "JAWS"
design; they had already decided to give up on the custom CPU and go with
the m68k. Sigh. Apparently three models were in development, with ICL
putting out the low-end uniprocessor "3a", and Pittsburgh joint-developing
the higher end "3b" and possibly multi-way "3c" machines. It's not clear,
of course, if "standard" m68k hardware with Accent+Qnix - or PNX - and the
fancy PERQ RasterOp engine would have survived against Sun either, but for
me the only "real" PERQs would have remained the bit-sliced boxes... I
mean, I like the m68k and all, but the microcoded CPU was totally badass!
:-] (April '83 Byte "Seive" benchmarks: PERQ is 7th fastest machine on
the list, blowing away all but the biggest Crays and IBM mainframes!)
Curator: Thanks for all this Chris! Correx made!